Intro
Contents
We have previously covered the Chinese audio brand Matrix Audio. We discussed the mini iPro 3. A nice little device for a favorable price. Matrix has not been sitting idle; serious devices have been added recently, including the Matrix SS-1 and SS-1 Pro switches. And how can your Network Nerd Author turn them down?
Matrix Audio is a manufacturer with a wide range of audio equipment. Its offerings range from power amplifiers to preamplifiers, streamers, transports and even switches. We test the two switches: the SS-1 and SS-1 Pro.
Construction
The Matrix SS-1 and SS-1 Pro look really neat. The housing is made of thick aluminum, which also makes the whole thing feel nice and sturdy. An added benefit is that this switch also weighs a fair bit and a thicker network cable won’t be able to pull it out of the rack.
On the back of the Pro, we see eight regular network ports and two SFP slots. With the regular SS-1, it’s six ports and two SFP slots. You can insert a fiber module into these SFP-slots, for example. We also tested that and it works without any problems.
With the Matrix SS-1 Pro, you can also connect an external 10 MHz clock. We connected the Mutec REF-10-120 to it and listened and measured it. Further down in the article you can read the results of the test.
Internally, we see a neat structure. The SS-1 Pro has a linear power supply, the regular SS-1 a switching model, with the option of using an external power supply. With the Matrix SS-1 Pro, we also see a slightly more luxurious clock and a different chipset; more fancy. That is necessary because of the two extra ports compared to the regular SS-1. Which chipset Matrix uses, we do not know; it is not a managed model and we are not going to remove heat-sinks, as we could potentially damage things.
I know at Alpha Audio, you’re not shy about self-criticism. So I’m curious—what led you to this thought: “Although the differences in our setup are subtle, perhaps we should start looking at a better streamer with a better internal clock to see if we might be able to get a little deeper with the measurements.”?
I find the tests you’re doing very insightful, and I wonder if you could run them with a high-end streamer or even a well-regarded, neutral-sounding streamer. Self-built solutions introduce so many variables—components, connections, cables, power supplies, etc.—and it feels like they don’t age as well in the sense that they aren’t updated as regularly as dedicated high-end streamers.
That said, I’m a bit confused—you mention the Volumio streamer, the AlphaPC/Streamer, and the Matrix MS-1 streamer. Are these all part of the tests, or are you considering other options? It would be interesting to understand how each of these fits into the evaluation process.
I would also love to see what happens if you setup a very focused streaming setup, with no added NAS or computer in the signal path. In my experience that is a bit of a game changer, if combined with good switcher and cables, and how many audiophiles are actually streaming. Any noise earlier in the signal chain seem to be very hard to completely eliminate so my experience is to try to remove everything and has a few power supplies (and good), as possible, on the ethernet signal path.
We have done that, extensively and in a large variety of options. Check our Streaming section in the menu, and you will see a plethora of reviews, live mulittests and interviews on streamers, switches, cables, power supplies and what not.
But I have a feeling you haven´t combined all the aspects at the same time? For example, using better quality ethernet cables (like AudioQuest Vodka for example) together with very good switches and power supplies (also earlier, on the router, and without a NAS connected). If you don´t target everything at the same time you are going to have weak links in your signal chain that will mask what ever you do later, in my humble experience, just using my ears though…
We are not obliged to test every variation that you can think of. Most of it is not interesting for the majority of our readers and that’s where our focus is.
There will be more speaker, amplifier and vinyl reviews coming, to counter balance what is becoming a worrying trend in our comments.
I think we have the most extensive tests on network equipment on the internet. If that is not enough for you… I don’t know what is. We cannot test every single combination. We test switches in a realistic environment. That is the most interesting for the majority or our readers. That is our audience…
The SS-1 Pro has the ability to disable rear port led lights. Your testing would have been a great time to see if this brought any measurable difference. Did you know it could do this?
Press rear button once to disable.
No, I did not know. Sorry.
I don’t really understand the use of a 10M reference clock in either Audio MCLK or here with Network equipment. Either way you have to go through a PLL to derive the frequencies you need. For network you need either a 25M or 50M oscillator depending on whose SWITCH chip you use. So really looking at 10M->25M or 10M->50M PLL is more of a challenge. Okay sure the PLL is suppose to inherit the jitter of the source clock. But really is that better than say a Crystek CCHD-950 at the root frequency?
How would you say the SS-1 PRO compares to the SOtM snh-10g SE with internal clock ect… they are priced similarly although you need an external PSU with the SOtM… but its also about 8 years old now, and the SS-1 PRO is brand new…
From Hans Beekhuyzen I beleive the snh-10g is of moderate value, I would bet these Matrix Audio are far superior, but I’m just speculating.
I haven’t compared them
Your measurements are very interesting Jaap, measuring the impact of those devices on jitter performance at the DAC is exactly what we want to see. Many thanks for doing that.
Also I was looking for a test that shows that a higher quality clock in a switch has an effect on jitter in the streamer/dac, you did it! Even if the listening results are mixed as you well described.
By doing all these tests you show more and more that the impact of such devices is real.
Thanks!
In my last real job before going full time with Wavelength I designed routers, repeaters, switches and bridges. The real test was data integrity and that would be the same here.
The effect on jitter at the dac level has to do with the amount of processing the endpoint has to do to complete a bit true stream of I2S. The more processing the heavier draw of power and the increased power will effect the MCLK performance.
But another thing with testing jitter is on time. Several years ago we were testing oscillators and it was a conclusion that devices that were swapped in and out (i.e. cold) would have a significant increase in jitter. We brought this up to several reviewers who would have their standard reference then drop in the review sample and not have positive results. For example a custom NDK 22.5792Mhz cold was 3.7pS, 15min 1.4pS, 1hr 5.7pS, 24hr 0.85pS. The phase noise graphs were really interesting as the cold had a large almost squaring up of 10dB between 8Hz and 20Hz. The 1hr was all over the place, but the 24hr was a nice smooth curve from -80dB at 1Hz to -140dB at 1Khz.
As I said in some other post here. One thing that kept bugging me was how an application could sound different. Audirvana sounds different than Roon, than JRiver etc… later I discovered it was due to how much processing. Roon was constantly phoning home with meta data transfers and so forth. That processing and network usage effect the power that everything saw and in tern effected the jitter performance and low level DAC chip performance. That increase in network traffic also can effect packet performance in a network setting. Everything makes a difference!
That’s a really interesting observation. So essentially, the test results show that an oscillator’s jitter and phase noise significantly improve as it warms up and stabilizes over time. A cold oscillator has much higher jitter, and the phase noise is more erratic, while after 24 hours, the jitter is minimized, and the phase noise curve smooths out.
This suggests that if someone swaps in a new device and evaluates it immediately, they might not be hearing or measuring its best performance. It makes sense that components with precision clocks, like network switches, could benefit from extended warm-up times to reach their optimal state.
Yes. Correct.
Interesting product and a good review, as always!
Unless i missed this information (i can´t find it), can you please share what ethernet cables you used since i don´t know if your using simple patch cables or AudioQuest Vodka (for example). That has an influence when you compare ethernet vs fiber, at least the differences gets smaller, in my experience
I used my own creation: triple shielded CAT7a with Telegartner plugs.
Thank you Jaap. And for the fiber connection, it would help to have it be between the SS-1 and the streamer?
Definitely! But only if the implementation is correct on the streamer side.
Question for the Network Nerd. There are a profusion of products that purport to clean up the ethernet signal prior to its entry in the streamer. There is the English Electric, the Muon Pro that Hans B. uses, etc. If I were to place the SS-1 switch prior to the streamer would it effectively do all the “cleanup” these other devices do? So I get the SS-1 and I’m done.
Yes
It’s great to read here about experiences with audiophile switches and PSU combinations! If a switch alone were enough to fully clean the signal, then “before the switch” upstream factors like router placement, power supplies, and network topology (such as using copper or fiber) wouldn’t have an audible impact—but real-world experience shows they do. Of course, the system needs to be revealing enough to make these differences noticeable.
Many users find improvements in sound quality by optimizing upstream elements such as power supplies, EMI/RFI shielding, and network isolation (e.g. router parameteres). This suggests that the router and overall network environment influence what reaches the switch. While the switch plays a key role, its performance is still shaped by the quality of the signal it receives. Fine-tuning these upstream factors allows the switch to perform at its best.
I have tested it and did not measure or hear a difference. But true: it is just one setup. And maybe it works in other environments.